Re: LL(1) vs LALR(1) parsers

elliottc@logica.com writes:
>LR parsers have many other advantages over LL(1) parsers....
>.... an LR parser can recognise a syntax error in the input
>stream as soon as it is possible to do so....


LL(k) parsers also have this ``viable prefix'' property.


By the way, I use a parser generator to generate recursive descent
LL(1) parsers. The parsers generated by my parser generator have
an even stronger property that Fischer and LeBlanc call the
``immediate error-detection'' property. That is, the parser
not only detects the error after reading the shortest possible
prefix of the input, it also detects the error before making any
internal state transitions. This allows for the best possible
error reporting and recovery.


To take advantage of this requires a lot of hand tuning, as
previously noted. But the parser generator guarantees that
you're starting with an LL(1) parser that is known to be correct
and is known to have the immediate error-detection property.


William D Clinger
--

Return to the comp.compilers page.
Search the comp.compilers archives again.